OP0115 GENERAL AND SEX-SPECIFIC PREDICTORS OF PSA AMONG PATIENTS WITH PSORIASIS

Abstract

Background: Risk prediction models in electronic health record (EHR) databases may assist in early identification of patients with psoriasis likely to develop psoriatic arthritis (PsA).1 A better understanding of potential predictors and whether stratification by sex would be needed in building such algorithms is required.1 Objectives: Examine general and sex-specific predictors of PsA in an EHR dataset among patients with psoriasis Methods: A retrospective cohort study was performed within the OptumInsights EHR Database (United States) between 2006-2017. Patients with two or more ICD codes for psoriasis and ages 16-90 were identified. The outcome was PsA (defined by a single ICD code). Potential predictors, in particular comorbidities and infections, were also identified using ICD codes. Hazard ratios were calculated using Cox proportional hazards models between individual predictors and development of incident PsA in univariate models and those that were significant (p<0.1) were entered into a multivariable model. A final model was achieved using automated stepwise regression. Separate models were developed for each sex as some predictors (e.g., polycystic ovarian syndrome, prostatitis) are sex-specific. Result(s): Among 215,386 patients with psoriasis, mean age was 50 (SD 15.6) and 55% were female. At index date (one year after date of first psoriasis code), 4.6% and 4.2% of patients had been prescribed a biologic therapy or oral therapy in the past year. Mean follow up time was 5.6 years (SD 2.8) and 4,288 patients developed incident PsA (incidence 3.5 cases/1,000 person years). Previously identified predictors were significant in univariate models (depression, fatigue, inflammatory bowel disease, uveitis, hyperlipidemia, fracture data not shown due to space restrictions) but several new predictors were also identified (diabetes, hidradenitis suppurativa, celiac disease, irritable bowel syndrome, sepsis, post-traumatic stress disorder, anxiety, anemia) (Table). Automated regression identified subsets of these factors in multivariable models these models differed by sex. Conclusion(s): Predictors of developing PsA differed by sex but obesity, depression, and fatigue were statistically significant predictors in both groups. Infections were also associated with development of PsA but the type of infection differed by sex.